PHYSIOLOGIA PLANTARUM 122: 181–189. 2004 Printed in Denmark – all rights reserved

doi: 10.1111/j.1399-3054.2004.00395.x Copyright # Physiologia Plantarum 2004

Different arabinogalactan proteins are present in carrot (Daucus carota) cell culture medium and in seeds Peter Immerzeela,b, Henk A. Scholsb, Alphons G. J. Voragenb and Sacco C. de Vriesa,* a

Laboratory of Biochemistry, Department of Agrotechnology and Food Sciences, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands b Laboratory of Food Chemistry, Department of Agrotechnology and Food Sciences, Wageningen University, Bomenweg 2, 6703 HD Wageningen, The Netherlands *Corresponding author, e-mail: [email protected] Received 29 March 2004; revised 11 June 2004

Arabinogalactan proteins (AGPs) were isolated by Yariv phenylglycoside precipitation from the medium of carrot (Daucus carota L.) cell cultures and from carrot seeds. The isolates showed a different composition of AGPs. The medium AGPs contained an arabinose poor AGP fraction that had relatively high levels of glucuronic acid and rhamnose. In contrast the seed AGPs only contained arabinose and galactose-rich AGP fractions that had low levels of glucuronic

acid. Linkage analysis on all fractions showed that most of the arabinose residues were terminally linked and that almost all galactose was present in the 1,3-, 1,6- and 1,3,6form. The strongly branched type II arabinogalactans are characteristic of the carbohydrate part of AGPs. AGP characteristic amino acid residues as Hyp, Pro, Glx, Ser, Gly, Asx, Ala, Leu and Thr were detected in three different fractions.

Introduction Arabinogalactan proteins (AGPs) are proteoglycans found in all higher plants that appear to be present in all stages of plant development and in all tissues investigated (Serpe and Nothnagel 1995, Gaspar et al. 2001). AGPs are found linked via glycosylphosphatidylinositol (GPI) anchors to the outer surface of the plasma membrane, bound to the cell wall or as secretions in intercellular spaces, culture media or the environment (Serpe and Nothnagel 1999, Showalter 2001). AGPs are found to be associated with processes as diverse as somatic embryogenesis, xylem development in primary roots, tip growth of pollen tubes and programmed cell death in cell cultures (Majewska-Sawka and Nothnagel 2000). Exogenously added AGPs or Yariv-mediated interference with AGP functioning often shows clear effects at the cellular level, however, the underlying molecular mechanisms of such AGP activities are still unsolved (Showalter 2001). Therefore, the precise biological functions of AGPs remain largely unknown.

AGPs belong to the hydroxyproline-rich glycoprotein (HRGP) family, that also includes the extensins, Pro/Hyprich glycoproteins (PRPs) and the solanaceaous lectins (Gaspar et al. 2001). The HRGP family comprises a continuum of molecules; from the-non- or minimally glycosylated PRPs, the moderately glycosylated extensins, through to the AGPs, where the carbohydrate moiety usually accounts for more than 90% by weight of the molecule (Sommer-Knudsen et al. 1998). Due to their high carbohydrate content AGPs are often referred to as proteoglycans (Sommer-Knudsen et al. 1998). Most AGPs can be specifically precipitated with Yariv phenylglycoside (Nothnagel 1997). However, some examples are known of molecules that have structural properties of AGPs but do not or hardly bind to Yariv phenylglycoside (Nothnagel 1997). The AGPs that have been investigated at the structural level have been isolated from the medium of cultured cells, plant exudates, xylem and stylar transmitting tract tissue (Serpe and Nothnagel 1995, Showalter

Abbreviations – AGP, arabinogalactan protein; EP3, extracellular protein 3; GPI, glycosylphosphatidylinositol; HRGP, hydroxyproline-rich glycoprotein; PRP, Pro/Hyp-rich glycoproteins. Physiol. Plant. 122, 2004

181

2001). The molecular size of soluble AGPs ranges from 60 to 300 kDa (Majewska-Sawka and Nothnagel 2000) although arabinogalactan-peptides are described by Fincher et al. (1974) that have an average molecular weight of 22 kDa. Mainly arabinose and galactose are present in the AGP carbohydrate side chains (Bacic et al. 1987). These are organized in type II arabinogalactans that contain a 1,3-b-D-galactopyranosyl backbone, 1,3,6b-D-galactopyranosyl branching points and 1,6-b-Dgalactopyranosyl side chains (Serpe and Nothnagel 1999, Showalter 2001). They carry a-arabino-furanosides as terminal residues and minor amounts of Glc, Rha, Xyl, Fuc and uronic acids (Baldwin et al. 1993). Type I arabinogalactans found in plants and micro-organisms differ from type II arabinogalactans in respect to their galactosyl residues that occur predominantly in 1,4-b-D-galactopyranosyl linkages (Serpe and Nothnagel 1999). The size of type II arabinogalactans that are O-linked to the protein core varies between 30 and 50 sugar residues (Serpe and Nothnagel 1999). Variation in the size, linkage and sequence of the arabinogalactans creates a wide chemical and structural diversity. Arabinose can also be linked to the protein core as small neutral oligosaccharides (Zhao et al. 2002). The proteins that are translated for classical AGPs contain a N-terminal secretion sequence that is not present on the mature protein, a central domain rich in Pro/Hyp and a C-terminal hydrophobic domain (Gaspar et al. 2001). According to Showalter (2001) AGPs should be classified as non-classical when the amino acid composition deviates from the classical AGPs, for instance when the protein core is Cys rich or Asn rich. Hydroxyprolinedeficient AGPs that have been isolated (Baldwin et al. 1993, Mollard and Joseleau 1994) should also be regarded as non-classical. The C-terminal hydrophobic domain functions as a signal for the attachment of a glycosylphosphatidylinositol (GPI) anchor (Gaspar et al. 2001). Classical C-terminal GPI anchor containing AGPs are found at the outer surface of the plasma membrane and are thought to interact with other molecules of the extracellular matrix (Zhao et al. 2002). Plant embryogenesis can be mimicked in vitro and was first described for suspension cultured carrot cells (Mordhorst et al. 1997). Cell lines produce a specific set of AGPs that are released into the medium and are able to stimulate the formation of embyos in carrot suspension cultures (Kreuger and van Holst 1993) and in Norway spruce (Egertsdotter and von Arnold 1995). With the use of crossed-electrophoresis it has been shown that the pattern of medium-isolated AGPs is related to the developmental stage of the cell culture (Kreuger and van Holst 1993). In addition to AGPs other carbohydrates and proteins are also present in the cell culture medium. de Jong et al. (1992) describe the extracellular protein 3 (EP3) endochitinase that was able to promote the formation of embryos of genetically modified carrot cultures at the non-permissive temperature. In previous experiments we have shown that carrot seed AGPs contained potential cleavage sites for the EP3 endochitinase. These results were based on the finding that both the EP3 endochitinase as well as seed AGPs were 182

able to increase the number of somatic embryos formed from protoplasts. Seed AGPs pre-treated with EP3 endochitinase were optimal in activity (van Hengel et al. 2001). It was also suggested that only a limited subset of AGPs contained such an endochitinase-sensitive carbohydrate moiety. To identify such a carbohydrate moiety and to increase our knowledge about the chemical composition of seed and medium AGPs, we fractionated AGPs from seeds and embryogenic cultures. The different fractions obtained were analysed on molecular size, chemical composition and some fractions were incubated with AGPdegrading enzymes. Our results suggest that AGPs from the different fractions are quite different in composition. The small amount of GlcN detected in the total seed AGP extract was, however, not retained in any of the further purified AGP fractions. This makes it unlikely that the previously proposed presence of GlcN in AGP preparations is an integral part of the AGP carbohydrate moiety.

Materials and methods Plant material Carrot (Daucus carota L. cv Autumn King/Trophy; Syngenta Seeds BV, Enkhuizen, The Netherlands) cell cultures were maintained as described in de Vries et al. 1988). One-week-old suspension cultures grown in 500 ml of Gamborg’s B5 medium, containing 0.2 mM 2,4-dichlorophenoxyacetic acid were used as a source for AGPs. Arabinogalactan protein isolation Suspension cells were removed by filtration and the medium was centrifuged at 13 900  g for 20 min. The AGPs from the supernatant were precipitated by adding 30 mg Yariv phenylglycoside to 1 l medium and NaCl at a final concentration of 0.15 M (Kreuger and van Holst 1993). The Yariv–AGP complex was precipitated at 4 C overnight, centrifuged at 13 900  g for 20 min, washed three times with 0.15 M NaCl and dissolved in water. Sodium hydrosulphite was added to decompose the b-glucosyl Yariv phenyl glycoside. The solution was heated to 50 C until the red colour disappeared, dialysed extensively against water at 4 C and freeze-dried. Carrot seeds (Daucus carota L. cv Yukon) were kindly provided by Dr Marc Kreuger (Syngenta Seeds BV, Enkhuizen, The Netherlands). Seeds (200 g) were ground in a cryo-mill to a fine powder that was re-suspended in 1 l distilled water at 4 C. After centrifugation at 13 900  g for 20 min the resulting pellet was extracted once with 1 l water at 4 C. The two extracts were pooled and the solubilized AGPs present in the extract were precipitated with 150 mg Yariv phenylglycoside. The Yariv–AGP complex was further processed as described for the medium AGPs. Sugar analysis Neutral sugar compositions were determined by gas chromatography according to Englyst and Cummings (1984). A solution of 0.1 mg AGPs was dried and hydrolysed in Physiol. Plant. 122, 2004

1 ml trifluoroacetic acid (TFA) (2 M) for 1 h at 121 C using inositol as internal standard. For the determination of glucosamine, allose was used as internal standard. After hydrolysis the released sugars were converted into their alditol acetates and analysed by GLC on a J & W (J&W Scientific Inc., Folsom, CA) DB-225 capillary column (15 m  0.53 mm, 1.0 mm film thickness) in a GC8000 Top gas chromatograph (Thermo Electron Corporation, Milan, Italy). The temperature programme was run from 60 to 180 C at 20 C min 1, 180 to 210 C at 2.5 C min 1 and at 210 C for 6 min. For detection a flame ionization detector was used. Confirmation of small amounts of GlcN was performed with a GC–MS using a SPB-1701 capillary column (30 m  0.32 mm, 0.25 mm film thickness; Supelco, Bellefonte, PA) in a gas chromatograph (HP 6890; Hewlett-Packard, Palo Alto, CA) coupled to a mass-selective detector (HP 5973; Hewlett-Packard) and using a HP Chem Station (Hewlett-Packard). The temperature program was run from 80 to 180 C at 20 C min 1, 180 to 250 C at 1.5 C min 1 and at 250 C for 3 min (van Hengel et al. 2001). The uronic acid content was determined by the automated colorimetric m-hydroxydiphenyl assay (Blumenkrantz and Asboe-Hansen 1973, Thibault 1979) using an auto-analyser (Skalar Analytical BV, Breda, The Netherlands). Galacturonic acid was used as a standard. To discriminate between glucuronic acid and galacturonic acid AGPs were subjected to methanolysis (de Ruiter et al. 1992). AGP aliquots (0.1 mg) were dried at 40 C under vacuum and hydrolysed with 2 M HCl in dry methanol for 16 h at 80 C. The monomeric sugars were analysed by High-Performance-Anion-Exchange-Chromatography (HPAEC) using a Thermo-separations HPLC system (Thermo Separations Products Inc., San Jose, CA) equipped with a Dionex ED40 Pulsed Amperometric Detection (Dionex Corporation, Sunnyvale, CA) with a Dionex CarboPac PA10 column and CarboPac PA-1 Guard column as described before (de Ruiter et al. 1992). The total sugar content is derived from data obtained from the different sugar residues present in the fractions as determined by GLC and HPAEC. Protein analysis Total protein content was determined by a semi-automated micro-Kjeldal method (Roozen and van Boxtel 1979). All nitrogen was assumed to originate from proteins and the conversion factor used was 6.25  N. Amino acids were analysed after hydrolysis of 5 mg AGPs with 6 M HCl at 110 C for 24 h using AOAC (1990) methodology (method no. 982.30). Separation was carried out on an amino acid analyser with a cation exchange column. Detection of the amino acids was carried out after post-column derivatization with ninhydrin at 440 and 570 nm. Cysteine, methionine and tryptophan were not included for the analysis. The determination of hydroxyproline was carried out according to the colorimetric method as described in Huszar et al. (1980). Physiol. Plant. 122, 2004

Sugar linkage analysis Methylation without a carboxyl reduction was performed with dimsyl anion and methyl iodide in DMSO (Sandford and Conrad 1966). AGPs were hydrolysed with 2 M TFA for 2 h and the monomers were converted into their partially methylated alditol acetates. The partially methylated alditol acetates were identified by GC–MS as described (Oosterveld et al. 2002). For quantification the samples were run on a GLC using flame ionisation detector (FID) detection. High-performance size-exclusion chromatography High-performance size-exclusion chromatography (HPSEC) was performed on three columns of Tosoh Bioscience TSK gel in series (G4000 PWXL, G3000 PWXL, G2500 PWXL), in combination with a PWXL-guard column (Tosoh Bioscience Inc., South San Francisco, CA). Elution took place at 30 C with 0.2 M NaNO3 at a flow rate of 0.8 ml min 1 (Kabel et al. 2002). The column effluent was monitored with a refractive index detector (Spectra System RI 150) (Thermo Electron Corporation, Milan, Italy). Pectins having molecular weight values in the range of 1000– 100 000 Da (as determined by viscometry) were used for calibration of the molecular weight. The neutral sugar percentage of the pectins was less than 10 mol%. Preparative size-exclusion chromatography Preparative size-exclusion chromatography was performed on a Sephacryl S-500 column (100  2.6 cm) using a Hiload System (Amersham Biotech) (Oosterveld et al. 2002). AGPs ( 110 mg) were applied to the column gel and eluted with 50 mM NaAc pH 5.0 at a flow rate of 2.5 ml min 1 and fractions were collected during 3 h over 180 tubes. The fractions were analysed for neutral sugar (Tollier and Robin 1979) and uronic acid (Blumenkrantz and AsboeHansen 1973, Thibault 1979) using an auto-analyser (Skalar Analytical BV) with arabinose and galacturonic acid as standards, respectively. Proteins were measured spectrophotometrically at 254 nm with a Shimadzu UV1601 spectrophotometer (Shimadzu, Tokyo, Japan)). Pooling the content of tubes together created three fractions. The formed fractions were dialysed and freeze-dried prior to further analysis. Radial Yariv gel diffusion A 1% agarose gel with 0.15 M NaCl and 0.03 g l 1 Yariv phenylglycoside was prepared according to van Holst and Clarke (1985). The wells were filled with AGPs (8 ml of 5 g l 1) and incubated overnight at room temperature in a box with high relative humidity to avoid drying of the gel. The Yariv phenylglycoside (1,3,5-tri [4-b-D-glucopyranosyl-oxyphenylazo]2,4,6-trihydroxybenzene) was prepared as described by Yariv et al. (1962). Enzymatic degradation The incubations were performed at 25 C for 16 h as described by Huisman et al. (1999) and references 183

Results Carrot AGPs were isolated with Yariv phenylglycoside from the medium of a non-embryogenic and an embryogenic cell culture and from commercially available seeds. The monosaccharide composition was largely the same for the AGPs from the embryogenic and the nonembryogenic culture (Table 1). Surprisingly, the total sugar content of the non-embryogenic culture was about 2.5 times lower than that of the embryogenic culture. The differences between the medium and seed AGPs is mainly caused by a lower amount of uronic acids in the seed AGPs. Arabinose and galactose are the most abundant sugar residues in all AGPs with Ara : Gal ratios of about 1 : 2. High arabinose and galactose content is indicative of AGPs (Sommer-Knudsen et al. 1998). Other neutral sugars are present in low amounts. Although glucosamine is not a common constituent of AGPs, minor amounts of amino sugars, especially GlcN, have been detected in several AGPs (Akiyama and Kato 1981, Serpe and Nothnagel 1995, van Hengel et al. 2001). Seed AGPs contain trace amounts of glucosamine (0.2 mol%), which was confirmed with GC–MSD. The presence of trace amounts of glucosamine in seed AGP preparations was confirmed in different cultivars and seed batches (data not shown). Glucosamine could not be detected in medium AGPs. All three AGP preparations were subjected to size exclusion chromatography to determine the size distribution.

Table 1. Monosaccharide composition (mol%) of medium AGPs and seed AGPs. NC, non-embryogenic cell culture; EC, embryogenic cell culture; tr, trace amounts (, 1%).

Figure 1 shows chromatograms of AGPs that were isolated from an embryogenic cell culture, a nonembryogenic cell culture and seeds. The SEC columns employed separate carbohydrates between approximately 200 and 0.2 kDa. The AGPs that were isolated from an embryogenic cell culture and the seed AGPs eluted in two distinct fractions. AGPs that were isolated from a non-embryogenic cell culture eluted in one peak (Mw about 45 kDa when compared with a pectin standard). Pectin was used for an indication of the size distribution as there are no AGP standards known. The interaction with the negatively charged column material and the negatively charged pectins will differ from the AGPs that have less charged sugar residues. Therefore the molecular weight of the AGPs might be somewhat underestimated. The material that elutes at 33 min is an unknown compound eluting at low Mw (approximately 400 Da) that shows absorption at a wavelength of 254 nm (data not shown). The ratio of the two populations of the embryogenic cell culture (100 and 45 kDa) can vary strongly among different embryogenic cell lines examined (data not shown). The elution profile of seed AGPs shows for both fractions a higher molecular weight; 105 and 75 kDa, respectively. The question arose whether both fractions indeed represent AGPs or other high molecular weight compounds. Since AGPs are known to aggregate in vitro (Sommer-Knudsen et al. 1998, Showalter 2001), we could also not exclude the possibility that the different fractions have the same AGP composition. Therefore, embryogenic culture AGPs and seed AGPs were separated on a semipreparative scale to allow a more detailed analysis of both fractions. The elution patterns of medium AGPs and seed AGPs again showed the presence of two high molecular weight fractions in Figs 2 and 3. The two high molecular weight fractions that were present using the analytical column were also base-line resolved using the preparative column and allowed us to

RI response

herein. AGPs at a concentration of 5 g l 1 in 0.05 M NaAc, pH 5.0 were incubated with b-galactosidase (0.22 mg ml 1 substrate solution) together with a-arabinofuranosidase (0.27 mg ml 1 substrate solution). Incubations were also performed with endo a-1,5-arabinanase (14.6 mg ml 1 substrate solution). b-galactosidase and a 1,5-arabinanase were purified from Aspergillus aculeatus, a-arabinofuranosidase was purified from Aspergillus niger. The amount of enzymes used was calculated to be sufficient for complete substrate digestion in 6 h. After the incubation the AGPs were heated for 10 min at 100 C to inactivate the enzymes. The AGP digests were analysed by HPSEC.

70

50

18

22

24

0.6

0.1

a

Medium AGPs

b

NC

EC

Seed AGPs

Rha Ara Xyl Man Gal Glc GlcN UA

5 23 0 0 50 4 0 19

5 24 tr tr 49 1 0 19

tr 34 tr tr 59 tr tr 4

Total sugar (w/w percentage)

32

81

75

184

110 90

c

16

20

26

28

30

32

34

36

38

Retention time Fig. 1. HPSEC elution profiles of an AGP extract from an embryogenic cell culture (a); a non-embryogenic cell culture (b); and of seed AGPs (c). Pectins were used as molecular weight standards (kDa). Physiol. Plant. 122, 2004

Fig. 2. Preparative SEC elution profiles of embryogenic cell culture medium AGPs on Sephacryl S-500. Thick line, neutral sugars; thin line, uronic acids; dashed line, UV254.

1

3000 AGPmI

AGPmIII

AGPmIV

2500

0.8

0.6 UV254

Conc. ug/ml

2000

1500 0.4 1000 0.2

500

0 0

50

100

150

200

250

300

350

400

0 450

Elution volume (mL)

pool the two different fractions. A fraction designated AGPmIV was pooled because it showed a high UV254 absorption, which might indicate that this fraction contains a higher protein concentration (Fig. 2; dashed line). In the seed AGPs a fraction designated AGPsII was pooled. This fraction contained material being eluted between the main peaks having an intermediate molecular mass. Table 2 shows the monosaccharide composition of the different fractions. For two fractions sufficient material was available for an additional total protein determination. AGPmI show low amounts of arabinose (3 mol%) resulting in an Ara : Gal ratio of 1 : 18. There is a high concentration of glucuronic acid in this fraction. AGPs with a high glucuronic acid content and low arabinose

concentration have been previously isolated from rose cell walls (Serpe and Nothnagel 1995). However, this is an unusual sugar composition for AGPs in which arabinose and galactose are the main carbohydrates as reported before (Nothnagel 1997). Fraction AGPmIII contains high amounts of arabinose and galactose and has an Ara : Gal ratio of 1 : 2.2. Rhamnose and glucuronic acid are present at lower levels. The total uronic acid content in all three medium fractions together is lower than the amount of uronic acids in the starting material (19 mol%, Table 1) as the main fraction (AGPmIII) contains 6 mol% GlcA. Uronic acids as shown in Table 1 were hydrolysed with sulphuric acid and the neutral sugars were hydrolysed with TFA, which is known to be not as strong as sulphuric acid. As the data of the UA

1400

5 AGPsI

AGPsII

AGPsIII

1200 4

3

800

UV254

Conc. ug/ml

1000

600

2

400 1 200 0 0

50

100

150

200

250

Elution volume (mL) Physiol. Plant. 122, 2004

300

350

400

0 450

Fig. 3. Preparative SEC elution profiles of seed AGPs on Sephacryl S-500. Thick line, neutral sugars; thin line, uronic acids; dashed line, UV254.

185

Table 2. Monosaccharide composition (mol%) of different fractions of embryogenic cell culture medium AGPs and seed AGPs obtained after size exclusion chromatography on Sephacryl S-500. AGPmI, AGPmIII, AGPmIV, medium AGPs, fractions I, III and IV respectively; AGPsI, AGPsII, AGPsIII, seed AGPs, fractions I, II and III respectively; tr, trace amounts (, 1%); ND, not determined. Medium AGPs

Seed AGPs

AGPmI

AGPmIII

AGPmIV

AGPsI

AGPsII

AGPsIII

Rha Ara Xyl Man Gal Glc GlcN GalA GlcA

11 3 5 5 56 0 0 tr 20

5 28 0 0 60 0 0 0 6

5 26 1 1 60 1 0 tr 6

1 37 0 tr 47 2 0 8 4

2 30 tr tr 59 3 0 tr 5

1 33 tr tr 61 0 0 tr 5

Total sugar (w/w percentage) Total protein (w/w percentage)

74 ND

89 2

62 ND

14 ND

36 ND

83 9

content and the data of the neutral sugars are merged in Table 1 the amount of uronic acids might be somewhat overestimated. The total amount of protein in this fraction is 2% (w/w). The sugar composition of AGPmIV is comparable with that of AGPmIII. Fraction AGPsI contains a very low amount of sugars (14% w/w) and shows a high UV254 absorption (Fig. 3). This might indicate a relatively high protein content for this fraction or implies incomplete dialysis. The main neutral sugars are arabinose and galactose, present in an Ara : Gal ratio of 1 : 1.3. Other neutral sugars are present in low amounts and galacturonic acid is present at 8 mol%. AGPsI differs from AGPsII and AGPsIII in the Ara : Gal ratio and the presence of galacturonic acid. AGPsII and AGPsIII show a comparable carbohydrate composition. Most of the dry weight material from the seed extracts is located in fraction AGPsIII. The total sugar content of AGPsIII is 83% (w/w) and the total protein content is 9% (w/w). Arabinose and galactose are the main contents in a ratio of 1 : 1.9. The low amount of glucosamine (0.2 mol%) that was detected in the crude Yariv extract of the seeds (Table 1) was not traceable in any of the three fractions. If GlcN was present in low molecular weight compounds they would not have been recovered here. The chemical composition of AGPmI is unusual for AGPs because of the low arabinose content and the relative high glucuronic acid content. In order to prove that all fractions contain AGPs, a Yariv radial diffusion gel was prepared to test the reactivity for all fractions. This gel showed for all fractions a precipitation with Yariv (Fig. 4), which is indicative for AGPs (Nothnagel 1997). For all fractions an identical amount of material was used. The smaller halo obtained for the first fractions of medium and seed AGPs might be caused by the higher molecular weight of the AGPs that impedes the diffusion through the gel. AGPmI, AGPmIII, AGPsI and AGPsIII were subjected to linkage analysis (Table 3). Arabinose was found to be present in the furanose form and occurs mainly as terminal sugar. In addition 1,5- linked arabinose was detected in all fractions at different concentrations. In seed AGPs 1,3,5-arabinose is detected in very small amounts and is completely absent in medium AGPs. Galactose is 186

detected in the pyranose form and is mainly present as 1,3-, 1,6- and 1,3,6- linked residues, unevenly distributed over all fractions. In all fractions galactose was also present as terminal residues, the highest amounts were observed in AGPmI. In medium AGPs trace amounts of galactose were found in the 1,4,6- form. Rhamnose and xylose were detected as terminal residues at high levels in AGPmI. These results show that the carbohydrates that are present in the AGP fractions are linked in a manner characteristic for type II arabinogalactans (Serpe and Nothnagel 1999). The ratio of terminal-linked sugar residues and branched sugar residues for AGPmI showed a high value of 2.6 and values approaching 1 for the other fractions. Amino acid analysis was performed on AGPmI, AGPmIII and on AGPsIII that were present in sufficient amounts. AGPmI contained many Pro, Glx, Ser, Gly, Asx and Ala residues (hydroxyproline not determined) whereas AGPmIII had high concentrations of Hyp, Pro, Ala, Gly, Ser and Leu (Table 4). AGPsIII had relatively high concentrations of Hyp, Ser, Ala, Glx, Gly and Thr residues. AGP-indicative amino acids such as Hyp, Pro,

AGPmI

AGPmIII

AGPmIV

AGPsI

AGPsII

AGPsIII

Fig. 4. Yariv phenylglycoside radial diffusion gel; AGPmI, AGPmIII, AGPmIV medium AGPs; AGPsI, AGPsII, AGPsIII seed AGPs. Physiol. Plant. 122, 2004

Table 3. Sugar linkage composition (mol%) of the main fractions of embryogenic culture medium AGPs and seed AGPs obtained after size exclusion chromatography on Sephacryl S-500. AGPmI, AGPmIII, medium AGPs, fractions I and III respectively; AGPsI, AGPsIII, seed AGPs, fractions I and III respectively; tr, trace amounts (, 1%). Medium AGPs Seed AGPs Residue

Linkage

Rhap Araf

terminal terminal 1,51,3,5unmethylated Xylp terminal Manp unmethylated Galp terminal 1,31,61,3,61,4,6unmethylated Glcp terminal 1,4Ratio t-linkages/branching points

AGPmI AGPmIII AGPsI AGPsIII 12 2 tr 0 0 11 tr 12 34 16 11 tr tr 0 tr 2.6

5 32 4 0 0 tr 0 5 8 12 34 tr tr 0 0 1.2

tr 33 6 tr tr 0 0 5 13 4 34 0 0 tr 1 1.1

tr 36 6 tr 0 0 tr 5 13 4 35 0 tr 0 0 1.1

Table 4. Amino acid composition (mol%) of AGPmI, AGPmIII, and AGPsIII. tr, trace amounts (, 1%); ND, not determined. Ornithine was not included in the table. Medium AGPs

Seed AGPs

Residue

AGPmI

AGPmIII

AGPsIII

Asx Thr Ser Glx Pro Gly Ala Val Ile Leu Tyr Phe Lys His Arg Hyp

9 4 9 10 15 9 7 5 4 5 3 4 4 4 3 ND

5 5 7 5 17 8 10 5 4 7 3 3 1 4 4 9

5 7 16 11 5 8 15 6 4 2 tr 2 3 5 2 9

Discussion

Physiol. Plant. 122, 2004

During this study AGPs were analysed that were isolated from medium and seeds. Medium AGPs from an embryogenic cell culture and seed AGPs could be separated into two main fractions that showed different molecular weight and chemical composition. In the different fractions it is not clear whether the protein core of the AGPs are identical or that different protein cores are

RI response

Medium AGPs

a b c 16

18

20

22

24

26

28

30

32

34

36

38

22

24

26

28

30

32

34

36

38

Seed AGPs RI response

Ala, Ser, Gly and Thr were present in the fractions at different concentrations. Two AGP fractions were incubated with galactosidase, arabinofuranosidase and endo arabinanase to determine whether the carbohydrate part of the AGPs could be used as substrate by the enzymes. Incubation of AGPmIII with galactosidase and arabinofuranosidase resulted in a slight decrease of the molecular weight and the formation of break down products (Fig. 5.1.b). The break down products elute between 33 and 37 min. The peak eluting at 35 min is caused by the buffer solution used for the enzymes. Incubation with arabinanase (Fig. 5.1.a) had no observable effects on this fraction. When seed AGPs (fraction III) were incubated with galactosidase and arabinofuranosidase a substantial part of the AGP aliquot is degraded (Fig. 5.2.b), even resulting in the formation of some oligomeric fragments. For seed AGPs no degradation was observed when this fraction was incubated with endo arabinanase (Fig. 5.2.a). The observation that the exo-enzymes were able to degrade the polysaccharides in an endo-fashion is probably due to very minor impurities of the enzymepreparations used. The enzyme incubation experiments showed that AGPs that have been extracted from seeds are more degradable with galactosidase and arabinofuranosidase when compared with medium AGPs. When medium and seed AGPs were incubated with a high concentration of the commercial protein hydrolysing enzyme (Pronase E) (Connolly et al. 1987) only the seed AGPs showed the formation of high molecular weight breakdown products (data not shown). The experiments with hydrolytic enzyme incubations indicate a high sensitivity for seed AGPs for both the carbohydrate part and the protein part of the molecule. Medium AGPs have a low substrate affinity for galactosidase and arabinofuranosidase.

a b c 16

18

20

Retention time

Fig. 5. HPSEC elution profiles of medium AGPs fraction III and of seed AGPs fraction III before (c) and after incubation with a 1-5 arabinanase (a) and b- galactosidase and a-arabinofuranosidase (b).

187

present. It is known for Arabidopsis for instance, that many different protein cores for AGPs are present (Schultz et al. 2000). In order to determine the glycosylation patterns of a specific protein, Zhao et al. (2002) labelled a specific AGP protein backbone from tomato with a fluorescent protein. These labelled AGPs could be traced in planta and allowed fractionation on the basis of the protein core. As we have no data available about the protein sequence of the AGPs, we are restricted to draw general conclusions about the obtained data of the isolates. The high molecular weight fraction AGPmI was found to contain low amounts of arabinose and high amounts of glucuronic acid and rhamnose. Serpe and Nothnagel (1995) found a cell wall AGP fraction in rose cell cultures with a low Ara : GlcA ratio and decided to designate this fraction as a glucuronogalactan protein (GGP). Rhamnose is present in high amounts in the same fraction. We can rule out the possibility that rhamnose is related to pectin as galacturonic acid is only present in trace amounts. All rhamnose residues are terminally linked to the carbohydrate chains. The 45 kDa fraction (AGPmIII) of the medium AGPs has high amounts of arabinose and galactose and low levels of rhamnose and glucuronic acid. The protein content is 2% (w/w) and this implies a remarkable small protein core for an AGP, even when the molecular weights of the AGPs are somewhat underestimated as pectins were used for the Mw standard. Therefore this fraction should be indicated as AG peptides (Schultz et al. 2000). From the monosaccharide composition data it can be concluded that the two main peaks present on HPSEC elution patterns contain macromolecules with a different carbohydrate composition. Therefore it is unlikely that a substantial part of the AGPs in the high molecular weight fraction (100 kDa) are aggregates of the AG peptides present in the low molecular weight fraction (45 kDa). The seed AGPs also contain two fractions and showed a higher molecular weight (105 and 75 kDa) when compared with the medium AGPs. The low molecular weight fraction has a protein content of 9% (w/w). In all seed fractions arabinose and galactose are the main sugars. The Ara : Gal ratio distinguishes the high and the low molecular weight fraction. Although present in the starting material, GlcN could not be detected in any of the fractions from the seed AGPs. It is not likely that the GlcN detected was part of a GPI-anchor. Although every anchor contains one GlcN residue, the AGPs with GPI-anchors also would have been detected after fractionation. N-linked glycans represent a family of glycans with a trimannosyl core and contain two contiguous GlcNAc residues. However, for the function of chitinase at least three contiguous GlcNAc residues are required. When the amino acid sequence is known, the occurrence of N-glycans can be predicted. To our knowledge no such sequences are known for carrot AGPs but we cannot exclude this possibility. Another possibility for the loss of GlcN detection is that GlcN was present in a low molecular weight compounds that were not pooled after fractionation. The oligosaccharides of 188

AGPs that changed in relative amount upon incubation with EP3 endochitinase (van Hengel et al. 2001) could be characterized in order to show that they contain glucosamine. Alternatively, the substrate activity of EP3 endochitinase should be studied in more detail. We can not rule out the possibility that the activity of EP3 endochitinase as observed on AGPs (van Hengel et al. 2001) is due to site activity of the enzyme. This means that, besides the ability of the enzyme to hydrolyse glycosidic bounds between GlcNAc, bounds between other sugar residues could also be hydrolysed. In that case the reduction in size of the AGPs after enzyme incubation is responsible for the increase in embryo formation after addition of enzyme-treated AGPs to carrot protoplasts. Linkage analysis data confirmed that all fractions contained type II arabinogalactans. However, the fraction AGPmI contained one third of the typical 1,3,6-b-D-Galp linkages typical for AGPs when compared with the other three fractions. Together with the high ratio of terminal linked sugar residues and branched sugar residues (2.6) we can conclude that this fraction contains less branched galactan chains and more unbranched oligosaccharides that are attached to the protein core. All fractions showed precipitation with Yariv phenylglycoside. The halo that was formed from the AGPmI fraction proved that these proteoglycans are still able to interact with Yariv despite the low arabinose concentration and the different galactose linkage composition. Hydroxyproline was detected in the two main AGP fractions from medium and seed. Other AGP characteristic amino acids were present in all fractions tested. Enzymatic degradation of carbohydrates and proteins showed that the seed AGPs were more degraded by hydrolytic enzymes than the AGPs that were isolated from the medium. This characteristic can be explained by a glycosylation pattern where both the protein part and the carbohydrate part are easy accessible for hydrolysing enzymes. Since the protein size of this seed AGP fraction is about seven times larger than the medium AG peptides, it is possible that the arabinogalactans are less closely packed on the seed-derived AGPs. Additional protein analysis could reveal data on the sequence of the proteins that are present in the AGP fractions. The hydroxyproline distribution of the protein cores and the corresponding attachment of the arabinogalactans depending on the amino acid sequence (Zhao et al. 2002) could prove this hypothesis. The amount of amino acids detected for AGPmIII is higher than the calculated total amount of amino acids in the protein core. Therefore it is likely that different protein cores are present in the obtained fractions. Additional research is required for a better insight of the complete structure of individual AGP molecules. A pre-requisite will be the isolation of AGP molecules with an identical protein backbone. Acknowledgements – The authors want to thank Dr M. Kreuger from Syngenta Seeds BV, Enkhuizen, The Netherlands for kindly providing the carrot seeds and R. Verhoef of the Laboratory of Physiol. Plant. 122, 2004

Food Chemistry, Department of Agrotechnology and Food Sciences of the Wageningen University for his help with the sugar linkage analysis.

References Akiyama Y, Kato K (1981) An extracellular arabinogalactan protein from Nicotiana tabacum. Phytochemistry 20: 2507–2510 AOAC (1990) Official Methods of Analysis. AOAC, Arlington, VA, pp 1096–1097 Bacic A, Churms SC, Stephen AM, Cohen PB, Fincher GB (1987) Fine structure of the arabinogalactan-protein from Lolium multiflorum. Carbohydr Res 162: 85–93 Baldwin TC, McCann MC, Roberts K (1993) A novel hydroxyprolinedeficient arabinogalactan protein secreted by suspension-cultured cells of Daucus carota. Plant Physiol 103: 115–123 Blumenkrantz N, Asboe-Hansen G (1973) New method for quantitative determination of uronic acids. Anal Biochem 54: 484–489 Connolly S, Fenyo J-C, Vandevelde M-C (1987) Heterogeneity and homogeneity of an arabinogalactan-protein: Acacia senegal gum. Food Hydrocolloids 5/6: 477–480 Egertsdotter U, von Arnold S, (1995) Importance of arabinogalactan proteins for the development of somatic embryos of Norway spruce (Picea abies). Physiol Plant 93: 334–345 Englyst HN, Cummings JH (1984) Simplified method the measurement of total non-starch polysaccharides by gas-liquid chromatography of constituent sugars as alditolacetates. Analyst 109: 937–942 Fincher GB, Sawyer WH, Stone BA (1974) Chemical and physical properties of an arabinogalactan-peptide from wheat endosperm. Biochem J 139: 535–545 Gaspar Y, Johnson KL, McKenna JA, Bacic A, Schultz CJ (2001) The complex structures of arabinogalactan-proteins and the journey towards understanding function. Plant Mol Biol 47: 161–176 van Hengel AJ, Tadesse Z, Immerzeel P, Schols H, van Kammen A, de Vries SC (2001) N-Acetylglucosamine and glucosaminecontaining arabinogalactan proteins control somatic embryogenesis. Plant Physiol 125: 1880–1890 van Holst G-J, Clarke AE (1985) Quantification of arabinogalactanprotein in plant extracts by single radial gel diffusion. Anal Biochem 148: 446–450 Huisman MMH, Schols HA, Voragen AGJ (1999) Enzymatic degradation of cell wall polysaccharides from soybean meal. Carbohydr Polym 38: 299–307 Huszar G, Maiocco J, Naftolin F (1980) Monitoring of collagen and collagen fragments in chromatography of protein mixtures. Anal Biochem 105: 424–429 de Jong AJ, Cordewener J, Schiavo FI, Terzi M, Vandekerckhove J, Van Kammen A, De Vries SC (1992) A carrot somatic embryo mutant is rescued by chitinase. Plant Cell 4: 425–433 Kabel MA, Schols HA, Voragen AGJ (2002) Complex xylo-oligosaccharides identified from hydro-thermally treated Eucalyptus wood and brewery’s spent grain. Carbohydr Polym 50: 191–200

Kreuger M, van Holst G-J (1993) Arabinogalactan proteins are essential in somatic embryogenesis of Daucus carota L. Planta 189: 243–248 Majewska-Sawka A, Nothnagel EA (2000) The multiple roles of arabinogalactan proteins in plant development. Plant Physiol 122: 3–9 Mollard A, Joseleau J-P (1994) Acacia senegal cells cultured in suspension secrete a hydroxyproline deficient arabinogalactanprotein. Plant Physiol Biochem 32: 703–709 Mordhorst AP, Toonen MAJ, de Vries SC (1997) Plant embryogenesis. Crit Rev Plant Sci 16: 535–576 Nothnagel EA (1997) Proteoglycans and related components in plant cells. Int Rev Cytol 174: 195–291 Oosterveld A, Voragen AGJ, Schols HA (2002) Characterization of hop pectins shows the presence of an arabinogalactan-protein. Carbohydr Polym 49: 407–413 Roozen JP, van Boxtel L (1979) Half-automatische bepaling van stikstof in levensmiddelen. De Ware(n)-Chemicus 9: 192–195 de Ruiter GA, Schols HA, Voragen AGJ, Rombouts FM (1992) Carbohydrate analysis of water-soluble uronic acid-containing polysaccharides with high-performance anion-exchange chromatography using methanolysis combined with TFA hydrolysis is superior to four other methods. Anal Biochem 207: 176–185 Sandford PA, Conrad HE (1966) The structure of the Aerobacter aerogenes A3 (S1) polysaccharide. I. A re-examination using improved procedures for methylation analysis. Biochemistry 5: 1508–1517 Schultz CJ, Johnson KL, Currie G, Bacic A (2000) The classical arabinogalactan protein gene family of Arabidopsis. Plant Cell 12: 1751–1767 Serpe MD, Nothnagel EA (1995) Fractionation and structural characterization of arabinogalactan-proteins from the cell wall of rose cells. Plant Physiol 109: 1007–1016 Serpe MD, Nothnagel EA (1999) Arabinogalactan-proteins in the multiple domains of the plant cell surface. Adv Bot Res 30: 207–289 Showalter AM (2001) Arabinogalactan-proteins: structure, expression and function. Cell Mol Life Sci 58: 1399–1417 Sommer-Knudsen J, Bacic A, Clarke AE (1998) Hydroxyprolinerich plant glycoproteins. Phytochemistry 47: 483–497 Thibault JF (1979) An automated method for the determination of pectic substances. Automatisation du dosage des substances pectiques par la me´thode au me´ta-hydroxydiphenyl. Lebensm– Wiss U Technol 12: 247–251 Tollier MT, Robin JP (1979) Adaptation de la me´thode a` l’orcinol sulfurique au dosage automatique des oses neutres et totaux. Ann Technol Agric 28: 1–15 de Vries SC, Booij H, Meyerink P, Huisman G, Wilde HD, Thomas TL, van Kammen A (1988) Acquisition of embryogenic potential in carrot cell-suspension cultures. Planta 176: 196–204 Yariv J, Rapport MM, Graf L (1962) The interaction of glycosides and saccharides with antibody to the corresponding phenylazo glycosides. Biochem J 85: 383–388 Zhao ZD, Tan L, Showalter AM, Lamport DTA, Kieliszewski MJ (2002) Tomato LeAGP-1 arabinogalactan-protein purified from transgenic tobacco corroborates the Hyp contiguity hypothesis. Plant J 31: 431–444

Edited by B. Sundberg Physiol. Plant. 122, 2004

189